Bias in Dynamic Panel Estimation with Fixed Effects, Incidental Trends and Cross Section Dependence

By Peter C. B. Phillips, Donggyu Sul. 2003

273 citations recorded by the source corpus.

Explicit asymptotic bias formulae are given for dynamic panel regression estimators as the cross section sample size N\rightarrow\infty. The results extend earlier work by Nickell (1981) in several directions that are relevant for practical work, including models with unit roots, deterministic trends, predetermined and exogenous regressors, and errors that may be cross sectionally dependent. The asymptotic bias is found to be so large when incidental linear trends are fitted and the time series sample size is small that it changes the sign of the autoregressive coefficient. Another finding of interest is that, when there is cross section error dependence, the probability limit of the dynamic panel regression estimator is a random variable rather than a constant, which helps to explain the substantial variability observed in dynamic panel estimates when there is cross section dependence even in situations where N is very large.

Subfields: Economics and Econometrics

View Paper: https://www.semanticscholar.org/paper/bdb3ea5e85f97bbd3715771683a0cfbc113c3343

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Economics and Econometrics

Citations in the corpus

Common Correlated Effects Estimation of Dynamic Panels with Cross-Sectional Dependence (2016, methodology): “Similar to the derivation of (A-20), (A-28) rewrites to plim N→∞ 1 NT N∑ i=1 y′i,−1MH i = plim N→∞ 1 NT N∑ i=1 T∑ t=1 yi,t−1 [ ( it − ̄i) − T∑ s=1 ̃+st ( is − ̄i) ] = − 1 T T∑ t=1 plim N→∞ 1 N N∑ i=1 yi,t−1̄i − 1T T∑ t=1 T∑ s=1 ̃+st plim N→∞ 1 N N∑ i=1 yi,t−1 is = − 2 T A( , T) − 2 1 T T−1∑ t=1 t−1g”
Bias-Corrected Estimation in Dynamic Panel Data Models (2005, background): “hence, as shown by Phillips and Sul (2004) already, the presence of cross-sectional heteroscedasticity does not have consequences for the particular form of the inconsistency.”
Smoothed quantile regression for panel data (2015, unknown): “To this end, as in Phillips and Sul (2007), we computed T times the bias of each estimator.”
Cross-Sectional Dependence in Panel Data Analysis (2012, methodology): “1̂FE = h 2" A ( ; T ) + (1) AT i h 2" B( ; T ) + (1) BT i 1 , (33) where A ( ; T ) and B( ; T ) are de ned in (30), and (1) AT = TX t=1 t 0 + 0 (wt 1 w; 1) , and (1) BT = TX t=1 (wt 1 w; 1)0 + 0 (wt 1 w; 1) , (34) 23See Nickell (1981) and Phillips and Sul (2007).”
Sieve estimation of panel data models with cross section dependence (2012, background): “See Bai (2009), Coakley et al. (2002), Greenaway-McGrevy et al. (2008), Harding (2007), Kapetanios and Pesaran (2005), Moon andWeidner (2008), Pesaran (2004, 2006), Pesaran and Tosetti (2007), Phillips and Sul (2003, 2007), among others, for an overview.”
Renewable and non-renewable energy consumption and economic activities: Further evidence from OECD countries (2014, background): “9 these problems conventional panel estimators (such as fixed or random effects) can result in misleading inference and even inconsistent estimators (Phillips and Sul, 2007).”
The Dark Side of Global Integration: Increasing Tail Dependence (2010, unknown): “Alternatives with higher disaggregation than the IMF Dummy include indicators by Quinn (1997) and Miniane (2004).”
Uncertainty, Financial Frictions, and Investment Dynamics (2014, methodology): “Finally, we explore this interaction using a newly constructed firm-level panel data set.”
Indirect Inference for Dynamic Panel Models (2010, unknown): “Using this formula and related formulae for cases with incidental trends (Phillips and Sul, 2007) it is easy to see that in many practically relevant cases the magnitude of the bias is considerable, and sometimes substantial enough to change the sign of the autoregressive coefficient estimate.”
Identifying Latent Structures in Panel Data (2016, background): “Phillips and Sul (2007b) provide explicit asymptotic bias formulae for linear dynamic panel regression estimators where the models may or may not exhibit unit roots, incidental trends, exogenous regressors, and cross section dependence, all of which lead to different formulae.”
THE ASYMPTOTIC PROPERTIES OF THE SYSTEM GMM ESTIMATOR IN DYNAMIC PANEL DATA MODELS WHEN BOTH N AND T ARE LARGE (2015, background): “These include Phillips and Sul (2003, 2007), Sarafidis and Robertson (2009), Sarafidis, Yamagata and Robertson (2009), Robertson, Sarafidis, and Symons (2010), Sarafidis and Yamagata (2010), Moon and Weidner (2010), and Hayakawa (2012).”
Asymptotic distribution of factor augmented estimators for panel regression (2012, background): “(This is shown by Phillips and Sul, 2007, p. 169, for the case when case for  =( 1)0.)”